QM9-XAS database of 56k QM9 small organic molecules labeled with TDDFT X-ray absorption spectra

Database for training graph neural network (GNN) models in Integrating Explainability into Graph Neural Network Models for the Prediction of X-ray Absorption Spectra, by Amir Kotobi, Kanishka Singh, Daniel Höche, Sadia Bari, Robert H.Meißner, and Annika Bande.

Included:

• qm9_Cedge_xas_56k.npz: the TDDFT XAS spectra of 56k structures from the QM9 dataset, were employed to label the graph dataset. The dataset contains two pairs of key/value entries: spec_stk, which represents a 2D array containing energies and oscillator strengths of XAS spectra, and id, which consists of the indices of QM9 structures. This data was used to create the QM9-XAS graph dataset.
• qm9xas_orca_output.zip: the raw ORCA output of TDDFT calculations for the 56k QM9-XAS dataset consists of excitation energies, densities, molecular orbitals, and other relevant information. This unprocessed output serves as a source to derive ground truth data for explaining the predictions made by GNNs.
• qm9xas_spec_train_val.pt: processed graph train/validation dataset of 50k QM9 structures. It is used as input to GNN models for training and validation.
• qm9xas_spec_test.pt: processed graph test dataset of 6k QM9 structures. It is used to test the performance of trained GNN models.

Notes on the datasets:

• The QM9-XAS dataset was created using ORCA electronic structure package [Neese, F., WIREs Computational Molecular Science 2012, 2, 73–78] to calculate carbon K-edge XAS spectra with the time-dependent density functional theory (TDDFT) method [Petersilka, M.; Gossmann, U. J.; Gross, E. K. U., Phys. Rev. Lett. 1996, 76, 1212–1215]
• The molecular structures of QM9-XAS datasets were sourced from the QM9 database [R. Ramakrishnan, P. O. Dral, M. Rupp, and O. A. Von Lilienfeld, Sci. Data 1, 1 (2014)].

Funding:

This research was funded by HIDA Trainee Network program, HAICU, Helmholtz AI-4-XAS, DASHH and HEIBRiDS graduate schools. For theoretical calculations and model training, computational resources at DESY and JFZ were used.